SILVER SPRING, MARYLAND—Every day, doctors have to figure out if they should dole out antibiotics or not. The decision is not only crucial to each patient’s wellbeing, but to public health in general. Unnecessary antibiotic doses help fuel drug-resistant infections that threaten tens of thousands of lives each year. Yet the question of whether someone actually needs an antibiotic is surprisingly hard to answer, leaving many doctors to decide based on hunches.
Last Friday, Ars sat in on a workshop held by the US Food and Drug Administration aimed at helping researchers and biotech companies develop new diagnostic tests that could take some of the guesswork out of the process. The idea was to start a conversation, medical officer Steven Gitterman of the FDA told Ars.
So far, one clear takeaway is that these tests are both desperately needed and technically challenging. In some cases, companies and the FDA may have to lower their standards regarding what makes an accurate diagnostic.
For years, researchers have worked on diagnostics to tell if a patient has a bacterial infection that can be zapped with antibiotics or perhaps a viral infection that can’t. Some of the tests aim to single out the specific microbe or virus that’s causing the illness. Others rely on complex signals from a patient’s immune response to reveal what kind of infection a person might be fighting.
But, so far, these tests have not poured out of the development pipeline and into the clinic. And that’s in part because of murky standards and expectations.
“In these kinds of developmental efforts, uncertainty is the enemy,” Sam Bozzette, a vice president at diagnostics company bioMérieux, told Ars. Among many uncertainties, companies and researchers have struggled to determine what makes a test good. Is it good if it’s correct 90 percent of the time? 70 percent? Even if we agreed on a standard, it’s difficult to compare how well new tests work because existing tests are unreliable, researchers said.
So far, surveying the nefarious germs hitchhiking on a sick patient has not been a reliable way to diagnose illness, researchers reported at the workshop. New tests are really good at spotting specific germs, but just because a person is carrying a potentially harmful bacterium doesn’t mean that it’s the one causing an infection—or that there’s even an infection and not something else driving the symptoms. Humans can harbor small groups of bad bugs without any consequences. Some can even quietly blend into our benign microbial communities.
That is why some people are looking at alternative approaches, like figuring out what the immune system is doing, infectious disease researcher Ephraim Tsalik of Duke University told Ars. He and colleagues are working up a new diagnostic test that probes dozens of genes that crank up or down depending on the type of infection the immune system is battling. Right now, the researchers are focusing on more than 100 genes, but they hope to trim that number to make a faster, more streamlined test, Tsalik said.
But such immune response-based tests will probably still not provide perfectly accurate, black and white results, Tsalik said. People’s immune responses differ depending on what other illnesses they have, their own personal genetics, the medications they take, and other such variables.
Luckily, perfect may not be necessary.
“We are just treating way too many people with antibiotics,” said panel member and infectious disease researcher Angela Caliendo of Brown University. Researchers need to do better at diagnosing infections, not find perfect, she said. “If we set a bar of 90 percent,” she said, “I don’t think we’ll ever get over it.”